The Export of Cryptography in the 20 Century and the 21
Total Page:16
File Type:pdf, Size:1020Kb
The Export of Cryptography th st in the 20 Century and the 21 ∗ Whitfield Diffie and Susan Landau Sun Microsystems, Inc Palo Alto CA April 19, 2005 August 2000 On the 14th of January 2000, the Bureau of Export Administration issued long-awaited revisions to the rules on exporting cryptographic hardware and software. The new regulations, which grew out of a protracted tug of war between the computer industry and the U.S. Government, are seen by indus- try as a victory. Their appearance, which was attended by both excitement and relief, marked a substantial change in export policy. This paper exam- ines the evolution of export control in the cryptographic area and considers its impact on the deployment of privacy-protecting technologies within the United States. Before the electronic age, all \real-time" interaction between people had to take place in person. Privacy in such interactions could be taken for granted. No more than reasonable care was required to assure yourself that only the people you were addressing | people who had to be right there with you | could hear you. Telecommunications have changed this. The people with whom you interact no longer have to be in your immediate vicinity; they can be on the other side of the world, making what was once impossible spontaneous and inexpensive. Telecommunication, on the other hand, makes protecting yourself from eavesdropping more difficult. Some other security mechanism is required to replace looking around to see that ∗To appear in the Handbook of the History of Information Security Elsevier B.V. 1 no one is close enough to overhear: that mechanism is cryptography, the only security mechanism that directly protects information passing out of the physical control of the sender and receiver. At the turn of the 20th century, cryptography was a labor-intensive, error- prone process incapable of more than transforming a small amount of written material into an encoded ciphertext form. At the dawn of the 21,st it can be done quickly, reliably, and inexpensively by computers at rates of a billion bits a second. This progress is commensurate with that of communications in general yet the fraction of the world's communications protected by cryptog- raphy today is still minuscule. In part this is due to the technical difficulty of integrating cryptography into communication systems so as to achieve se- curity, in part to an associated marketing problem. Proper implementation of cryptosecurity requires substantial up-front expenditure on infrastructure while most of the benefit is lost unless there is nearly ubiquitous coverage, a combination that deters investment. These factors result in a lack of robust- ness of the market that makes it prey to a third factor, political opposition. As telecommunication has improved in quality and gained in importance, police and intelligence organizations have made ever more extensive use of the possibilities for electronic eavesdropping. These same agencies now fear that the growth of cryptography in the commercial world will deprive them of sources of information on which they have come to rely. The result has been a struggle between the business community, which needs cryptography to protect electronic commerce and elements of government that fear the loss of their surveillance capabilities. Export control has emerged as an important battleground in this struggle. Background In the 1970s, after many years as the virtually exclusive property of the military, cryptography appeared in public with a dual thrust. First came the work of Horst Feistel and others at IBM that produced the U.S. Data En- cryption Standard. DES, which was adopted in 1977 as Federal Information Processing Standard 46, was mandated for the protection of all government information legally requiring protection but not covered under the provisions for protecting classified information | a category later called “unclassified sensitive." The second development was the work of several academics that was to lead to public-key cryptography, the technology underlying the security of 2 internet commerce today. Public-key cryptography makes it possible for two people, without having arranged a secret key in advance, to communicate securely over an insecure channel. Public-key cryptography also provides a digital signature mechanism remarkably similar in function to a written signature.1 The effect of new developments in distinct areas of cryptography was to ignite a storm of interest in the field, leading to an explosion of papers, books, and conferences. The government response was to try to acquire the same sort of \born classified” legal control over cryptography that the Department of Energy claimed2 in the area of atomic energy. The effort was a dramatic failure. NSA hoped an American Council on Education committee set up to study the problem would recommend legal restraints on cryptographic research and publication. Instead, it proposed only that authors voluntary submit papers to NSA for its opinion on the possible national-security implications of their publication.3 It did not take the government long to realize that even if control of research and publication were beyond its grasp, control of deployment was not. Although laws directly regulating the use of cryptography in the U.S. appeared out of reach | and no serious effort was ever made to get Congress to adopt any | adroit use of export control proved remarkably effective in diminishing the use of cryptography, not only outside the United States but inside as well. 1In recognition of the increasing importance of electronic commerce, in June 2000, President Clinton signed into law the Millennium Digital Commerce Act, Public Law 106- 229, which establishes the legal validity of \electronic signatures." The term is somewhat broader than \digital signature" but points the way toward a future in which alternatives to the written signature play a central role in commerce. 2The courts have never ruled on the constitutionality of this provision of the Atomic Energy Act of 1946. U.S. Statutes at Large 60 (1947); 755-75. In 1997, the Progressive magazine challenged it by proposing to publish an article by Robert Morland entitled \The H-bomb Secret, how we got it, why we're telling it." After the appearance of an independent and far less competent article on how h-bombs work, the government succeeded in having the case mooted and leaving the impression in the popular mind that it would have won. In fact, the virtual certainty that it would have lost is undoubtedly why it acted as it did. 3Susan Landau, Primes, Codes and the National Security Agency, 30 Notices of the American Mathematical Society 7, at 10 (1983). 3 Export Control The export control laws in force today are rooted in the growth of the Cold War that followed World War II. In the immediate post-war years the U.S. accounted for a little more than half of the world's economy. Furthermore, the country was just coming off a war footing, with its machinery of production controls, rationing, censorship, and economic warfare. The U.S. thus had not only the economic power to make export control an effective element of foreign policy but the inclination and the regulatory machinery to do so. The system that grew out of this environment had not one export control regime but two. Primary legal authority for regulating exports was given to the Department of State, with the objective of protecting national security. Although the goods to be regulated are described as munitions, the law does not limit itself to the common meaning of that word and includes many things that are neither explosive nor dangerous. The affected items are determined by the Department of State acting, through the Munition Control Board, on the advice of other elements of the executive branch, especially, in the case of cryptography, the National Security Agency.4 Exports that are deemed to have civilian as well as military uses are reg- ulated by the Department of Commerce. Such items are termed dual-use and present a wholly different problem from \munitions." A broad range of goods | vehicles, aircraft, clothing, copying machines - - | are vital to military functioning just as they are to civilian. If the sale of such goods was routinely blocked merely because they might benefit the military of an unfriendly country, there would be little left of international trade. Control of the export of dual-use articles therefore balances considerations of mili- tary application with considerations of foreign availability | the existence of sources of supply prepared to fill any vacuum left by U.S. export bans. The munitions controls are far more severe than the dual-use controls, requiring individually approved export licenses specifying the product and the actual customer as opposed to broad restrictions by product category and national destination. Legal authority to decide which regime is to be applied lies with the Department of State, which can authorize the transfer of jurisdiction to the Department of Commerce, a process called commodities jurisdiction. Assessing whether a product is military or civilian is not always straight- 4U.S. Export control regulations are described at length in Root and Liebman's United States Export Controls.[Root] 4 forward. Once we leave the domain of the clearly military (such as fighter aircraft), we immediately encounter products that either have both military and civilian uses or products that can be converted from one to the other without difficulty. The Boeing 707, a civilian airplane, was a mainstay of the world's airlines during the 1960s and 1970s. Its military derivatives, the C-135 (cargo, including passengers), the KC-135 (tanker), and RC-135 (intelligence platform) have been mainstays of Western military aviation. Recognizing that civilian aircraft might be put to military use and thus by- pass export control, the U.S.